test/html-webhacc/cc.cgi

#!/usr/bin/perl
use strict;

use lib qw[/home/httpd/html/www/markup/html/whatpm
           /home/wakaba/work/manakai/lib
           /home/wakaba/public_html/-temp/wiki/lib];
use CGI::Carp qw[fatalsToBrowser];
use Scalar::Util qw[refaddr];

use SuikaWiki::Input::HTTP; ## TODO: Use some better CGI module

sub htescape ($) {
  my $s = $_[0];
  $s =~ s/&/&amp;/g;
  $s =~ s/</&lt;/g;
  $s =~ s/>/&gt;/g;
  $s =~ s/"/&quot;/g;
  $s =~ s!([\x00-\x09\x0B-\x1F\x7F-\x80])!sprintf '<var>U+%04X</var>', ord $1!ge;
  return $s;
} # htescape

my $http = SuikaWiki::Input::HTTP->new;

## TODO: _charset_

  if ($http->meta_variable ('PATH_INFO') ne '/') {
    print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400";
    exit;
  }

  my $input_format = $http->parameter ('i') || 'text/html';
  my $inner_html_element = $http->parameter ('e');
  my $input_uri = 'thismessage:/';

  my $s = $http->parameter ('s');
  if (length $s > 1000_000) {
    print STDOUT "Status: 400 Document Too Long\nContent-Type: text/plain; charset=us-ascii\n\nToo long";
    exit;
  }

  load_text_catalog ('en'); ## TODO: conneg

  my @nav;
  print STDOUT qq[Content-Type: text/html; charset=utf-8

<!DOCTYPE html>
<html lang="en">
<head>
<title>Web Document Conformance Checker (BETA)</title>
<link rel="stylesheet" href="../cc-style.css" type="text/css">
</head>
<body>
<h1>Web Document Conformance Checker (<em>beta</em>)</h1>

<div id="document-info" class="section">
<dl>
<dt>Document URI</dt>
    <dd><code class="URI" lang="">&lt;<a href="@{[htescape $input_uri]}">@{[htescape $input_uri]}</a>&gt;</code></dd>
<dt>Internet Media Type</dt>
    <dd><code class="MIME" lang="en">@{[htescape $input_format]}</code></dd>
]; # no </dl> yet
  push @nav, ['#document-info' => 'Information'];

  require Message::DOM::DOMImplementation;
  my $dom = Message::DOM::DOMImplementation->____new;
  my $doc;
  my $el;

  if ($input_format eq 'text/html') {
    require Encode;
    require Whatpm::HTML;
    
    $s = Encode::decode ('utf-8', $s);

    print STDOUT qq[
<dt>Character Encoding</dt>
    <dd>(none)</dd>
</dl>
</div>

<div id="source-string" class="section">
<h2>Document Source</h2>
];
    push @nav, ['#source-string' => 'Source'];
    print_source_string (\$s);
    print STDOUT qq[
</div>

<div id="parse-errors" class="section">
<h2>Parse Errors</h2>

<dl>
];
  push @nav, ['#parse-errors' => 'Parse Error'];

  my $onerror = sub {
    my (%opt) = @_;
    my ($cls, $msg) = get_text ($opt{type}, $opt{level});
    if ($opt{column} > 0) {
      print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a> column $opt{column}</dt>\n];
    } else {
      $opt{line} = $opt{line} - 1 || 1;
      print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a></dt>\n];
    }
    $opt{type} =~ tr/ /-/;
    $opt{type} =~ s/\|/%7C/g;
    $msg .= qq[ [<a href="../error-description#$opt{type}">Description</a>]];
    print STDOUT qq[<dd class="$cls">$msg</dd>\n];
  };

  $doc = $dom->create_document;
  if (defined $inner_html_element and length $inner_html_element) {
    $el = $doc->create_element_ns
        ('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]);
    Whatpm::HTML->set_inner_html ($el, $s, $onerror);
  } else {
    Whatpm::HTML->parse_string ($s => $doc, $onerror);
  }

  print STDOUT qq[
</dl>
</div>
];
  } elsif ($input_format eq 'application/xhtml+xml') {
    require Message::DOM::XMLParserTemp;
    require Encode;
    
    my $t = Encode::decode ('utf-8', $s);

    print STDOUT qq[
<dt>Character Encoding</dt>
    <dd>(none)</dd>
</dl>
</div>

<div id="source-string" class="section">
<h2>Document Source</h2>
];
    push @nav, ['#source-string' => 'Source'];
    print_source_string (\$t);
    print STDOUT qq[
</div>

<div id="parse-errors" class="section">
<h2>Parse Errors</h2>

<dl>];
  push @nav, ['#parse-errors' => 'Parse Error'];

  my $onerror = sub {
    my $err = shift;
    my $line = $err->location->line_number;
    print STDOUT qq[<dt><a href="#line-$line">Line $line</a> column ];
    print STDOUT $err->location->column_number, "</dt><dd>";
    print STDOUT htescape $err->text, "</dd>\n";
    return 1;
  };

  open my $fh, '<', \$s;
  $doc = Message::DOM::XMLParserTemp->parse_byte_stream
      ($fh => $dom, $onerror, charset => 'utf-8');

    print STDOUT qq[</dl>
</div>
];
  } else {
    print STDOUT qq[
</dl>
</div>

<div id="result-summary" class="section">
<p><em>Media type <code class="MIME" lang="en">@{[htescape $input_format]}</code> is not supported!</em></p>
</div>
];
    push @nav, ['#result-summary' => 'Result'];
  }


  if (defined $doc or defined $el) {
    print STDOUT qq[
<div id="document-tree" class="section">
<h2>Document Tree</h2>
];
    push @nav, ['#document-tree' => 'Tree'];

    print_document_tree ($el || $doc);

    print STDOUT qq[
</div>

<div id="document-errors" class="section">
<h2>Document Errors</h2>

<dl>];
    push @nav, ['#document-errors' => 'Document Error'];

    require Whatpm::ContentChecker;
    my $onerror = sub {
      my %opt = @_;
      my ($cls, $msg) = get_text ($opt{type}, $opt{level});
      $opt{type} = $opt{level} . ':' . $opt{type} if defined $opt{level};
      $opt{type} =~ tr/ /-/;
      $opt{type} =~ s/\|/%7C/g;
      $msg .= qq[ [<a href="../error-description#$opt{type}">Description</a>]];
      print STDOUT qq[<dt class="$cls">] . get_node_link ($opt{node}) .
          qq[</dt>\n<dd class="$cls">], $msg, "</dd>\n";
    };

    my $elements;
    if ($el) {
      $elements = Whatpm::ContentChecker->check_element ($el, $onerror);
    } else {
      $elements = Whatpm::ContentChecker->check_document ($doc, $onerror);
    }

    print STDOUT qq[</dl>
</div>
];

    if (@{$elements->{table}}) {
      require JSON;

      print STDOUT qq[
<div id="tables" class="section">
<h2>Tables</h2>

<!--[if IE]><script type="text/javascript" src="../excanvas.js"></script><![endif]-->
<script src="../table-script.js" type="text/javascript"></script>
<noscript>
<p><em>Structure of tables are visualized here if scripting is enabled.</em></p>
</noscript>
];

      my $i = 0;
      for my $table_el (@{$elements->{table}}) {
        $i++;
        print STDOUT qq[<div class="section" id="table-$i"><h3>] .
            get_node_link ($table_el) . q[</h3>];
        
        my $table = Whatpm::HTMLTable->form_table ($table_el);
        
        for (@{$table->{column_group}}, @{$table->{column}}, $table->{caption}) {
          next unless $_;
          delete $_->{element};
        }
        
        for (@{$table->{row_group}}) {
          next unless $_;
          next unless $_->{element};
          $_->{type} = $_->{element}->manakai_local_name;
          delete $_->{element};
        }
        
        for (@{$table->{cell}}) {
          next unless $_;
          for (@{$_}) {
            next unless $_;
            for (@$_) {
              $_->{id} = refaddr $_->{element} if defined $_->{element};
              delete $_->{element};
            }
          }
        }
        
        print STDOUT '</div><script type="text/javascript">tableToCanvas (';
        print STDOUT JSON::objToJson ($table);
        print STDOUT qq[, document.getElementById ('table-$i'));</script>];
      }
    
      print STDOUT qq[</div>];
    }

    if (keys %{$elements->{term}}) {
      print STDOUT qq[
<div id="terms" class="section">
<h2>Terms</h2>

<dl>
];
      for my $term (sort {$a cmp $b} keys %{$elements->{term}}) {
        print STDOUT qq[<dt>@{[htescape $term]}</dt>];
        for (@{$elements->{term}->{$term}}) {
          print STDOUT qq[<dd>].get_node_link ($_).qq[</dd>];
        }
      }
      print STDOUT qq[</dl></div>];
    }
  }

  ## TODO: Show result

  print STDOUT qq[
<ul class="navigation" id="nav-items">
];
  for (@nav) {
    print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>];
  }
  print STDOUT qq[
</ul>
</body>
</html>
];

exit;

sub print_source_string ($) {
  my $s = $_[0];
  my $i = 1;
  print STDOUT qq[<ol lang="">\n];
  if (length $$s) {
    while ($$s =~ /\G([^\x0A]*?)\x0D?\x0A/gc) {
      print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n";
      $i++;
    }
    if ($$s =~ /\G([^\x0A]+)/gc) {
      print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n";
    }
  } else {
    print STDOUT q[<li id="line-1"></li>];
  }
  print STDOUT "</ol>";
} # print_input_string

sub print_document_tree ($) {
  my $node = shift;
  my $r = '<ol class="xoxo">';

  my @node = ($node);
  while (@node) {
    my $child = shift @node;
    unless (ref $child) {
      $r .= $child;
      next;
    }

    my $node_id = 'node-'.refaddr $child;
    my $nt = $child->node_type;
    if ($nt == $child->ELEMENT_NODE) {
      my $child_nsuri = $child->namespace_uri;
      $r .= qq[<li id="$node_id" class="tree-element"><code title="@{[defined $child_nsuri ? $child_nsuri : '']}">] . htescape ($child->tag_name) .
          '</code>'; ## ISSUE: case

      if ($child->has_attributes) {
        $r .= '<ul class="attributes">';
        for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value, $_->namespace_uri, 'node-'.refaddr $_] }
                      @{$child->attributes}) {
          $r .= qq[<li id="$attr->[3]" class="tree-attribute"><code title="@{[defined $_->[2] ? $_->[2] : '']}">] . htescape ($attr->[0]) . '</code> = '; ## ISSUE: case?
          $r .= '<q>' . htescape ($attr->[1]) . '</q></li>'; ## TODO: children
        }
        $r .= '</ul>';
      }

      if ($child->has_child_nodes) {
        $r .= '<ol class="children">';
        unshift @node, @{$child->child_nodes}, '</ol></li>';
      } else {
        $r .= '</li>';
      }
    } elsif ($nt == $child->TEXT_NODE) {
      $r .= qq'<li id="$node_id" class="tree-text"><q lang="">' . htescape ($child->data) . '</q></li>';
    } elsif ($nt == $child->CDATA_SECTION_NODE) {
      $r .= qq'<li id="$node_id" class="tree-cdata"><code>&lt;[CDATA[</code><q lang="">' . htescape ($child->data) . '</q><code>]]&gt;</code></li>';
    } elsif ($nt == $child->COMMENT_NODE) {
      $r .= qq'<li id="$node_id" class="tree-comment"><code>&lt;!--</code><q lang="">' . htescape ($child->data) . '</q><code>--&gt;</code></li>';
    } elsif ($nt == $child->DOCUMENT_NODE) {
      $r .= qq'<li id="$node_id" class="tree-document">Document';
      $r .= qq[<ul class="attributes">];
      $r .= qq[<li>@{[scalar get_text ('manakaiIsHTML:'.($child->manakai_is_html?1:0))]}</li>];
      $r .= qq[<li>@{[scalar get_text ('manakaiCompatMode:'.$child->manakai_compat_mode)]}</li>];
      $r .= qq[</ul>];
      if ($child->has_child_nodes) {
        $r .= '<ol class="children">';
        unshift @node, @{$child->child_nodes}, '</ol></li>';
      }
    } elsif ($nt == $child->DOCUMENT_TYPE_NODE) {
      $r .= qq'<li id="$node_id" class="tree-doctype"><code>&lt;!DOCTYPE&gt;</code><ul class="attributes">';
      $r .= qq[<li class="tree-doctype-name">Name = <q>@{[htescape ($child->name)]}</q></li>];
      $r .= qq[<li class="tree-doctype-publicid">Public identifier = <q>@{[htescape ($child->public_id)]}</q></li>];
      $r .= qq[<li class="tree-doctype-systemid">System identifier = <q>@{[htescape ($child->system_id)]}</q></li>];
      $r .= '</ul></li>';
    } elsif ($nt == $child->PROCESSING_INSTRUCTION_NODE) {
      $r .= qq'<li id="$node_id" class="tree-id"><code>&lt;?@{[htescape ($child->target)]}</code> <q>@{[htescape ($child->data)]}</q><code>?&gt;</code></li>';
    } else {
      $r .= qq'<li id="$node_id" class="tree-unknown">@{[$child->node_type]} @{[htescape ($child->node_name)]}</li>'; # error
    }
  }

  $r .= '</ol>';
  print STDOUT $r;
} # print_document_tree

sub get_node_path ($) {
  my $node = shift;
  my @r;
  while (defined $node) {
    my $rs;
    if ($node->node_type == 1) {
      $rs = $node->manakai_local_name;
      $node = $node->parent_node;
    } elsif ($node->node_type == 2) {
      $rs = '@' . $node->manakai_local_name;
      $node = $node->owner_element;
    } elsif ($node->node_type == 3) {
      $rs = '"' . $node->data . '"';
      $node = $node->parent_node;
    } elsif ($node->node_type == 9) {
      $rs = '';
      $node = $node->parent_node;
    } else {
      $rs = '#' . $node->node_type;
      $node = $node->parent_node;
    }
    unshift @r, $rs;
  }
  return join '/', @r;
} # get_node_path

sub get_node_link ($) {
  return qq[<a href="#node-@{[refaddr $_[0]]}">] .
      htescape (get_node_path ($_[0])) . qq[</a>];
} # get_node_link

{
  my $Msg = {};

sub load_text_catalog ($) {
  my $lang = shift; # MUST be a canonical lang name
  open my $file, '<', "cc-msg.$lang.txt" or die "$0: cc-msg.$lang.txt: $!";
  while (<$file>) {
    if (s/^([^;]+);([^;]*);//) {
      my ($type, $cls, $msg) = ($1, $2, $_);
      $msg =~ tr/\x0D\x0A//d;
      $Msg->{$type} = [$cls, $msg];
    }
  }
} # load_text_catalog

sub get_text ($) {
  my ($type, $level) = @_;
  $type = $level . ':' . $type if defined $level;
  my @arg;
  {
    if (defined $Msg->{$type}) {
      my $msg = $Msg->{$type}->[1];
      $msg =~ s/\$([0-9]+)/defined $arg[$1] ? htescape ($arg[$1]) : '(undef)'/ge;
      return ($Msg->{$type}->[0], $msg);
    } elsif ($type =~ s/:([^:]*)$//) {
      unshift @arg, $1;
      redo;
    }
  }
  return ('', htescape ($_[0]));
} # get_text

}

=head1 AUTHOR

Wakaba <w@suika.fam.cx>.

=head1 LICENSE

Copyright 2007 Wakaba <w@suika.fam.cx>

This library is free software; you can redistribute it
and/or modify it under the same terms as Perl itself.

=cut

## $Date: 2007/07/01 06:21:46 $
1	wakaba	1.1	#!/usr/bin/perl
2			use strict;
3
4			use lib qw[/home/httpd/html/www/markup/html/whatpm
5			/home/wakaba/work/manakai/lib
6			/home/wakaba/public_html/-temp/wiki/lib];
7			use CGI::Carp qw[fatalsToBrowser];
8	wakaba	1.2	use Scalar::Util qw[refaddr];
9	wakaba	1.1
10			use SuikaWiki::Input::HTTP; ## TODO: Use some better CGI module
11
12	wakaba	1.2	sub htescape ($) {
13			my $s = $_[0];
14			$s =~ s/&/&/g;
15			$s =~ s/</</g;
16			$s =~ s/>/>/g;
17			$s =~ s/"/"/g;
18			$s =~ s!([\x00-\x09\x0B-\x1F\x7F-\x80])!sprintf '<var>U+%04X</var>', ord $1!ge;
19			return $s;
20			} # htescape
21
22	wakaba	1.1	my $http = SuikaWiki::Input::HTTP->new;
23
24			## TODO: _charset_
25
26	wakaba	1.8	if ($http->meta_variable ('PATH_INFO') ne '/') {
27			print STDOUT "Status: 404 Not Found\nContent-Type: text/plain; charset=us-ascii\n\n400";
28			exit;
29			}
30
31	wakaba	1.2	my $input_format = $http->parameter ('i') \|\| 'text/html';
32			my $inner_html_element = $http->parameter ('e');
33			my $input_uri = 'thismessage:/';
34	wakaba	1.1
35			my $s = $http->parameter ('s');
36			if (length $s > 1000_000) {
37			print STDOUT "Status: 400 Document Too Long\nContent-Type: text/plain; charset=us-ascii\n\nToo long";
38			exit;
39			}
40	wakaba	1.2
41	wakaba	1.7	load_text_catalog ('en'); ## TODO: conneg
42
43	wakaba	1.3	my @nav;
44	wakaba	1.2	print STDOUT qq[Content-Type: text/html; charset=utf-8
45
46			<!DOCTYPE html>
47			<html lang="en">
48			<head>
49			<title>Web Document Conformance Checker (BETA)</title>
50	wakaba	1.3	<link rel="stylesheet" href="../cc-style.css" type="text/css">
51	wakaba	1.2	</head>
52			<body>
53			<h1>Web Document Conformance Checker (<em>beta</em>)</h1>
54
55	wakaba	1.4	<div id="document-info" class="section">
56	wakaba	1.2	<dl>
57			<dt>Document URI</dt>
58			<dd><code class="URI" lang=""><<a href="@{[htescape $input_uri]}">@{[htescape $input_uri]}</a>></code></dd>
59			<dt>Internet Media Type</dt>
60			<dd><code class="MIME" lang="en">@{[htescape $input_format]}</code></dd>
61			]; # no </dl> yet
62	wakaba	1.3	push @nav, ['#document-info' => 'Information'];
63	wakaba	1.1
64			require Message::DOM::DOMImplementation;
65			my $dom = Message::DOM::DOMImplementation->____new;
66			my $doc;
67			my $el;
68
69	wakaba	1.2	if ($input_format eq 'text/html') {
70			require Encode;
71			require Whatpm::HTML;
72
73			$s = Encode::decode ('utf-8', $s);
74
75			print STDOUT qq[
76			<dt>Character Encoding</dt>
77			<dd>(none)</dd>
78			</dl>
79	wakaba	1.3	</div>
80	wakaba	1.2
81			<div id="source-string" class="section">
82	wakaba	1.3	<h2>Document Source</h2>
83	wakaba	1.2	];
84	wakaba	1.3	push @nav, ['#source-string' => 'Source'];
85	wakaba	1.2	print_source_string (\$s);
86			print STDOUT qq[
87			</div>
88	wakaba	1.1
89	wakaba	1.2	<div id="parse-errors" class="section">
90			<h2>Parse Errors</h2>
91
92	wakaba	1.4	<dl>
93	wakaba	1.2	];
94	wakaba	1.3	push @nav, ['#parse-errors' => 'Parse Error'];
95	wakaba	1.1
96			my $onerror = sub {
97			my (%opt) = @_;
98	wakaba	1.7	my ($cls, $msg) = get_text ($opt{type}, $opt{level});
99	wakaba	1.2	if ($opt{column} > 0) {
100	wakaba	1.7	print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a> column $opt{column}</dt>\n];
101	wakaba	1.2	} else {
102	wakaba	1.7	$opt{line} = $opt{line} - 1 \|\| 1;
103			print STDOUT qq[<dt class="$cls"><a href="#line-$opt{line}">Line $opt{line}</a></dt>\n];
104	wakaba	1.2	}
105	wakaba	1.8	$opt{type} =~ tr/ /-/;
106			$opt{type} =~ s/\\|/%7C/g;
107			$msg .= qq[ [<a href="../error-description#$opt{type}">Description</a>]];
108	wakaba	1.7	print STDOUT qq[<dd class="$cls">$msg</dd>\n];
109	wakaba	1.1	};
110
111			$doc = $dom->create_document;
112	wakaba	1.2	if (defined $inner_html_element and length $inner_html_element) {
113	wakaba	1.1	$el = $doc->create_element_ns
114	wakaba	1.2	('http://www.w3.org/1999/xhtml', [undef, $inner_html_element]);
115	wakaba	1.1	Whatpm::HTML->set_inner_html ($el, $s, $onerror);
116			} else {
117			Whatpm::HTML->parse_string ($s => $doc, $onerror);
118			}
119
120	wakaba	1.2	print STDOUT qq[
121	wakaba	1.4	</dl>
122	wakaba	1.2	</div>
123			];
124			} elsif ($input_format eq 'application/xhtml+xml') {
125			require Message::DOM::XMLParserTemp;
126			require Encode;
127
128			my $t = Encode::decode ('utf-8', $s);
129
130			print STDOUT qq[
131			<dt>Character Encoding</dt>
132			<dd>(none)</dd>
133			</dl>
134	wakaba	1.3	</div>
135	wakaba	1.2
136			<div id="source-string" class="section">
137	wakaba	1.3	<h2>Document Source</h2>
138	wakaba	1.2	];
139	wakaba	1.3	push @nav, ['#source-string' => 'Source'];
140	wakaba	1.2	print_source_string (\$t);
141			print STDOUT qq[
142			</div>
143	wakaba	1.1
144	wakaba	1.2	<div id="parse-errors" class="section">
145			<h2>Parse Errors</h2>
146	wakaba	1.1
147	wakaba	1.7	<dl>];
148	wakaba	1.3	push @nav, ['#parse-errors' => 'Parse Error'];
149	wakaba	1.1
150			my $onerror = sub {
151			my $err = shift;
152	wakaba	1.2	my $line = $err->location->line_number;
153	wakaba	1.4	print STDOUT qq[<dt><a href="#line-$line">Line $line</a> column ];
154			print STDOUT $err->location->column_number, "</dt><dd>";
155			print STDOUT htescape $err->text, "</dd>\n";
156	wakaba	1.1	return 1;
157			};
158
159			open my $fh, '<', \$s;
160			$doc = Message::DOM::XMLParserTemp->parse_byte_stream
161			($fh => $dom, $onerror, charset => 'utf-8');
162
163	wakaba	1.7	print STDOUT qq[</dl>
164	wakaba	1.2	</div>
165			];
166			} else {
167			print STDOUT qq[
168			</dl>
169	wakaba	1.4	</div>
170	wakaba	1.2
171	wakaba	1.3	<div id="result-summary" class="section">
172	wakaba	1.2	<p><em>Media type <code class="MIME" lang="en">@{[htescape $input_format]}</code> is not supported!</em></p>
173	wakaba	1.3	</div>
174	wakaba	1.2	];
175	wakaba	1.3	push @nav, ['#result-summary' => 'Result'];
176	wakaba	1.2	}
177
178
179			if (defined $doc or defined $el) {
180			print STDOUT qq[
181			<div id="document-tree" class="section">
182			<h2>Document Tree</h2>
183			];
184	wakaba	1.3	push @nav, ['#document-tree' => 'Tree'];
185	wakaba	1.2
186			print_document_tree ($el \|\| $doc);
187
188			print STDOUT qq[
189			</div>
190	wakaba	1.1
191	wakaba	1.2	<div id="document-errors" class="section">
192			<h2>Document Errors</h2>
193
194	wakaba	1.7	<dl>];
195	wakaba	1.3	push @nav, ['#document-errors' => 'Document Error'];
196	wakaba	1.1
197			require Whatpm::ContentChecker;
198			my $onerror = sub {
199			my %opt = @_;
200	wakaba	1.7	my ($cls, $msg) = get_text ($opt{type}, $opt{level});
201	wakaba	1.8	$opt{type} = $opt{level} . ':' . $opt{type} if defined $opt{level};
202			$opt{type} =~ tr/ /-/;
203			$opt{type} =~ s/\\|/%7C/g;
204			$msg .= qq[ [<a href="../error-description#$opt{type}">Description</a>]];
205	wakaba	1.7	print STDOUT qq[<dt class="$cls">] . get_node_link ($opt{node}) .
206			qq[</dt>\n<dd class="$cls">], $msg, "</dd>\n";
207	wakaba	1.1	};
208	wakaba	1.2
209	wakaba	1.5	my $elements;
210	wakaba	1.1	if ($el) {
211	wakaba	1.5	$elements = Whatpm::ContentChecker->check_element ($el, $onerror);
212	wakaba	1.1	} else {
213	wakaba	1.5	$elements = Whatpm::ContentChecker->check_document ($doc, $onerror);
214	wakaba	1.1	}
215	wakaba	1.2
216	wakaba	1.7	print STDOUT qq[</dl>
217	wakaba	1.2	</div>
218			];
219	wakaba	1.5
220			if (@{$elements->{table}}) {
221			require JSON;
222
223			print STDOUT qq[
224			<div id="tables" class="section">
225			<h2>Tables</h2>
226
227			<!--[if IE]><script type="text/javascript" src="../excanvas.js"></script><![endif]-->
228			<script src="../table-script.js" type="text/javascript"></script>
229			<noscript>
230			<p><em>Structure of tables are visualized here if scripting is enabled.</em></p>
231			</noscript>
232			];
233
234			my $i = 0;
235			for my $table_el (@{$elements->{table}}) {
236			$i++;
237	wakaba	1.6	print STDOUT qq[<div class="section" id="table-$i"><h3>] .
238			get_node_link ($table_el) . q[</h3>];
239	wakaba	1.5
240			my $table = Whatpm::HTMLTable->form_table ($table_el);
241
242			for (@{$table->{column_group}}, @{$table->{column}}, $table->{caption}) {
243			next unless $_;
244			delete $_->{element};
245			}
246
247			for (@{$table->{row_group}}) {
248			next unless $_;
249			next unless $_->{element};
250			$_->{type} = $_->{element}->manakai_local_name;
251			delete $_->{element};
252			}
253
254			for (@{$table->{cell}}) {
255			next unless $_;
256			for (@{$_}) {
257			next unless $_;
258			for (@$_) {
259			$_->{id} = refaddr $_->{element} if defined $_->{element};
260			delete $_->{element};
261			}
262			}
263			}
264
265			print STDOUT '</div><script type="text/javascript">tableToCanvas (';
266			print STDOUT JSON::objToJson ($table);
267			print STDOUT qq[, document.getElementById ('table-$i'));</script>];
268			}
269
270			print STDOUT qq[</div>];
271			}
272	wakaba	1.6
273			if (keys %{$elements->{term}}) {
274			print STDOUT qq[
275			<div id="terms" class="section">
276			<h2>Terms</h2>
277
278			<dl>
279			];
280			for my $term (sort {$a cmp $b} keys %{$elements->{term}}) {
281			print STDOUT qq[<dt>@{[htescape $term]}</dt>];
282			for (@{$elements->{term}->{$term}}) {
283			print STDOUT qq[<dd>].get_node_link ($_).qq[</dd>];
284			}
285			}
286			print STDOUT qq[</dl></div>];
287			}
288	wakaba	1.1	}
289
290	wakaba	1.2	## TODO: Show result
291	wakaba	1.3
292	wakaba	1.2	print STDOUT qq[
293	wakaba	1.3	<ul class="navigation" id="nav-items">
294			];
295			for (@nav) {
296			print STDOUT qq[<li><a href="$_->[0]">$_->[1]</a></li>];
297			}
298			print STDOUT qq[
299			</ul>
300	wakaba	1.2	</body>
301			</html>
302			];
303	wakaba	1.1
304			exit;
305
306	wakaba	1.2	sub print_source_string ($) {
307			my $s = $_[0];
308			my $i = 1;
309			print STDOUT qq[<ol lang="">\n];
310	wakaba	1.7	if (length $$s) {
311			while ($$s =~ /\G([^\x0A]*?)\x0D?\x0A/gc) {
312			print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n";
313			$i++;
314			}
315			if ($$s =~ /\G([^\x0A]+)/gc) {
316			print STDOUT qq[<li id="line-$i">], htescape $1, "</li>\n";
317			}
318			} else {
319			print STDOUT q[<li id="line-1"></li>];
320	wakaba	1.2	}
321			print STDOUT "</ol>";
322			} # print_input_string
323
324			sub print_document_tree ($) {
325	wakaba	1.1	my $node = shift;
326	wakaba	1.2	my $r = '<ol class="xoxo">';
327	wakaba	1.1
328	wakaba	1.2	my @node = ($node);
329	wakaba	1.1	while (@node) {
330			my $child = shift @node;
331	wakaba	1.2	unless (ref $child) {
332			$r .= $child;
333			next;
334			}
335
336			my $node_id = 'node-'.refaddr $child;
337			my $nt = $child->node_type;
338			if ($nt == $child->ELEMENT_NODE) {
339	wakaba	1.4	my $child_nsuri = $child->namespace_uri;
340			$r .= qq[<li id="$node_id" class="tree-element"><code title="@{[defined $child_nsuri ? $child_nsuri : '']}">] . htescape ($child->tag_name) .
341	wakaba	1.2	'</code>'; ## ISSUE: case
342
343			if ($child->has_attributes) {
344			$r .= '<ul class="attributes">';
345	wakaba	1.4	for my $attr (sort {$a->[0] cmp $b->[0]} map { [$_->name, $_->value, $_->namespace_uri, 'node-'.refaddr $_] }
346	wakaba	1.2	@{$child->attributes}) {
347	wakaba	1.4	$r .= qq[<li id="$attr->[3]" class="tree-attribute"><code title="@{[defined $_->[2] ? $_->[2] : '']}">] . htescape ($attr->[0]) . '</code> = '; ## ISSUE: case?
348	wakaba	1.2	$r .= '<q>' . htescape ($attr->[1]) . '</q></li>'; ## TODO: children
349			}
350			$r .= '</ul>';
351			}
352
353	wakaba	1.7	if ($child->has_child_nodes) {
354	wakaba	1.2	$r .= '<ol class="children">';
355	wakaba	1.6	unshift @node, @{$child->child_nodes}, '</ol></li>';
356			} else {
357			$r .= '</li>';
358	wakaba	1.2	}
359			} elsif ($nt == $child->TEXT_NODE) {
360	wakaba	1.4	$r .= qq'<li id="$node_id" class="tree-text"><q lang="">' . htescape ($child->data) . '</q></li>';
361	wakaba	1.2	} elsif ($nt == $child->CDATA_SECTION_NODE) {
362	wakaba	1.4	$r .= qq'<li id="$node_id" class="tree-cdata"><code><[CDATA[</code><q lang="">' . htescape ($child->data) . '</q><code>]]></code></li>';
363	wakaba	1.2	} elsif ($nt == $child->COMMENT_NODE) {
364	wakaba	1.4	$r .= qq'<li id="$node_id" class="tree-comment"><code><!--</code><q lang="">' . htescape ($child->data) . '</q><code>--></code></li>';
365	wakaba	1.2	} elsif ($nt == $child->DOCUMENT_NODE) {
366	wakaba	1.6	$r .= qq'<li id="$node_id" class="tree-document">Document';
367	wakaba	1.7	$r .= qq[<ul class="attributes">];
368			$r .= qq[<li>@{[scalar get_text ('manakaiIsHTML:'.($child->manakai_is_html?1:0))]}</li>];
369			$r .= qq[<li>@{[scalar get_text ('manakaiCompatMode:'.$child->manakai_compat_mode)]}</li>];
370			$r .= qq[</ul>];
371	wakaba	1.2	if ($child->has_child_nodes) {
372	wakaba	1.7	$r .= '<ol class="children">';
373	wakaba	1.6	unshift @node, @{$child->child_nodes}, '</ol></li>';
374	wakaba	1.1	}
375	wakaba	1.2	} elsif ($nt == $child->DOCUMENT_TYPE_NODE) {
376	wakaba	1.5	$r .= qq'<li id="$node_id" class="tree-doctype"><code><!DOCTYPE></code><ul class="attributes">';
377			$r .= qq[<li class="tree-doctype-name">Name = <q>@{[htescape ($child->name)]}</q></li>];
378			$r .= qq[<li class="tree-doctype-publicid">Public identifier = <q>@{[htescape ($child->public_id)]}</q></li>];
379			$r .= qq[<li class="tree-doctype-systemid">System identifier = <q>@{[htescape ($child->system_id)]}</q></li>];
380	wakaba	1.2	$r .= '</ul></li>';
381			} elsif ($nt == $child->PROCESSING_INSTRUCTION_NODE) {
382	wakaba	1.4	$r .= qq'<li id="$node_id" class="tree-id"><code><?@{[htescape ($child->target)]}</code> <q>@{[htescape ($child->data)]}</q><code>?></code></li>';
383	wakaba	1.1	} else {
384	wakaba	1.4	$r .= qq'<li id="$node_id" class="tree-unknown">@{[$child->node_type]} @{[htescape ($child->node_name)]}</li>'; # error
385	wakaba	1.1	}
386			}
387	wakaba	1.2
388			$r .= '</ol>';
389			print STDOUT $r;
390			} # print_document_tree
391	wakaba	1.1
392			sub get_node_path ($) {
393			my $node = shift;
394			my @r;
395			while (defined $node) {
396			my $rs;
397			if ($node->node_type == 1) {
398			$rs = $node->manakai_local_name;
399			$node = $node->parent_node;
400			} elsif ($node->node_type == 2) {
401			$rs = '@' . $node->manakai_local_name;
402			$node = $node->owner_element;
403			} elsif ($node->node_type == 3) {
404			$rs = '"' . $node->data . '"';
405			$node = $node->parent_node;
406			} elsif ($node->node_type == 9) {
407			$rs = '';
408			$node = $node->parent_node;
409			} else {
410			$rs = '#' . $node->node_type;
411			$node = $node->parent_node;
412			}
413			unshift @r, $rs;
414			}
415			return join '/', @r;
416			} # get_node_path
417
418	wakaba	1.6	sub get_node_link ($) {
419			return qq[<a href="#node-@{[refaddr $_[0]]}">] .
420			htescape (get_node_path ($_[0])) . qq[</a>];
421			} # get_node_link
422
423	wakaba	1.7	{
424			my $Msg = {};
425
426			sub load_text_catalog ($) {
427			my $lang = shift; # MUST be a canonical lang name
428			open my $file, '<', "cc-msg.$lang.txt" or die "$0: cc-msg.$lang.txt: $!";
429			while (<$file>) {
430			if (s/^([^;]+);([^;]*);//) {
431			my ($type, $cls, $msg) = ($1, $2, $_);
432			$msg =~ tr/\x0D\x0A//d;
433			$Msg->{$type} = [$cls, $msg];
434			}
435			}
436			} # load_text_catalog
437
438			sub get_text ($) {
439			my ($type, $level) = @_;
440			$type = $level . ':' . $type if defined $level;
441			my @arg;
442			{
443			if (defined $Msg->{$type}) {
444			my $msg = $Msg->{$type}->[1];
445			$msg =~ s/\$([0-9]+)/defined $arg[$1] ? htescape ($arg[$1]) : '(undef)'/ge;
446			return ($Msg->{$type}->[0], $msg);
447			} elsif ($type =~ s/:([^:]*)$//) {
448			unshift @arg, $1;
449			redo;
450			}
451			}
452			return ('', htescape ($_[0]));
453			} # get_text
454
455			}
456
457	wakaba	1.1	=head1 AUTHOR
458
459			Wakaba <w@suika.fam.cx>.
460
461			=head1 LICENSE
462
463			Copyright 2007 Wakaba <w@suika.fam.cx>
464
465			This library is free software; you can redistribute it
466			and/or modify it under the same terms as Perl itself.
467
468			=cut
469
470	wakaba	1.8	## $Date: 2007/07/01 06:21:46 $